Complement associated microvascular injury and in the pathogenesis of severe COVID-19 infection: A report of five cases

CYNTHIA MAGRO, J. JUSTIN MULVEY, DAVID BERLIN, GERARD NUOVO, STEVEN SALVATORE, JOANNA HARP, AMELIA BAXTER-STOLTZFUS, and JEFFREY LAURENCE

NEW YORK, NEW YORK; POWELL, OHIO; AND NEW YORK, NEW YORKXX

Acute respiratory failure and a systemic are critical aspects of the morbidity and mortality characterizing infection with severe acute respiratory dis- tress syndrome-associated coronavirus-2, the etiologic agent of Coronavirus dis- ease 2019 (COVID-19). We examined skin and tissues from 5 patients with severe COVID-19 characterized by respiratory failure (n= 5) and purpuric skin (n = 3). COVID-19 pneumonitis was predominantly a pauci-inflammatory septal injury with significant septal capillary mural and luminal fibrin deposition and permeation of the interalveolar septa by neutrophils. No viral cytopathic changes were observed and the diffuse alveolar damage (DAD) with hyaline mem- branes, inflammation, and type II pneumocyte hyperplasia, hallmarks of classic acute respiratory distress syndrome, were not prominent. These pulmonary findings were accompanied by significant deposits of terminal complement components C5b-9 (membrane attack complex), C4d, and mannose binding lectin (MBL)-asso- ciated serine protease (MASP)2, in the microvasculature, consistent with sustained, systemic activation of the complement pathways. The purpuric skin similarly showed a pauci-inflammatory thrombogenic vasculopathy, with deposition of C5b- 9 and C4d in both grossly involved and normally-appearing skin. In addition, there was co-localization of COVID-19 spike glycoproteins with C4d and C5b-9 in the interalveolar septa and the cutaneous microvasculature of 2 cases examined. In conclusion, at least a subset of sustained, severe COVID-19 may define a type of catastrophic microvascular injury syndrome mediated by activation of comple- ment pathways and an associated procoagulant state. It provides a foundation for further exploration of the pathophysiologic importance of complement in COVID- 19, and could suggest targets for specific intervention. (Translational Research 2020; 220:1À13)

From the Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, New York; Department of Laboratory Medicine, Memorial Sloan-Kettering Cancer Center, New York, New York; Department of Medicine, Division of Pulmonary and Critical Care Medicine, Weill Cornell Medicine, New York, New York; The Ohio State University Comprehensive Cancer Center, Columbus Ohio and Discovery Life Sci- ences, Powell, Ohio; Department of Dermatology, Weill Cornell Medicine; Department of Medicine, Division of Hematology and Medical Oncol- ogy, Weill Cornell Medicine, New York, New York. Submitted for PublicationApril 9, 2020; accepted for publication April 9, 2020. Reprint requests: Jeffrey Laurence, Weill Cornell Medicine, 1300 York Avenue, New York, NY 10065. E-mail address: [email protected]. 1931-5244/$ - see front matter Ó 2020 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.trsl.2020.04.007

1 Translational Research 2 Magro et al June 2020

of acute respiratory distress syndrome (ARDS).7 Pre- AT A GLANCE COMMENTARY liminary pathology studies of COVID-19 patients dem- onstrated diffuse alveolar damage (DAD) with , Background hyaline membranes, and inflammation, followed by The respiratory distress syndrome accompanying type II pneumocyte hyperplasia, features characteristic 8,9 a subset of severe Coronavirus disease 2019 of typical ARDS. But many patients with COVID- (COVID-19) may be distinct from classic acute 19-related severe respiratory distress have a delayed 10 respiratory distress syndrome. There is relatively onset of respiratory distress, then manifest relatively well-preserved lung mechanics despite the sever- well-preserved lung mechanics, despite the severity of ity of hypoxemia, characterized by high respira- hypoxemia, characterized by high respiratory compli- tory compliance and high shunt fraction, and ance and high shunt fraction, and prolonged require- 10,11 increasing recognition of systemic features of a ment for mechanical ventilation. Therefore, hypercoaguable state in this disease. Therefore, significant aspects of the pathology of COVID-19 the pathology and pathophysiology of COVID-19 might be expected to differ from classic ARDS. might differ from that of typical acute respiratory We examined lung and cutaneous tissues from distress syndrome. We sought to define the role of 5 patients with SARS-CoV-2 infection and severe complement activation and microvascular throm- respiratory failure, 3 of whom also had features con- bosis in cases of persistent, severe COVID-19. sistent with a systemic procoagulant state, including retiform purpura or livedo racemosa—prominent Translational Significance dermatologic signs of a generalized microvascular thrombotic disorder—and markedly elevated A pattern of tissue damage consistent with com- d-dimers. Histologic and immunohistochemistry plement-mediated microvascular injury was noted studies defined a pattern of cutaneous and pulmonary in the lung and/or skin of 5 individuals with severe pathology involving microvascular injury and throm- COVID-19. Our demonstration of the striking bosis, consistent with activation of the alternative deposition of C5b-9, C4d, and MASP2 in the pathway (AP) and lectin pathway (LP) of comple- microvasculature of 2 organ systems is consistent ment. Co-localization of SARS-CoV-2-specific spike with profound and generalized activation of both alternative and lectin-based pathways. It provides glycoproteins with complement components in the lung and skin was also documented. Our studies sug- a foundation for further exploration of the patho- gest that at least a subset of severe COVID-19 infec- physiologic importance of complement in tion involves a catastrophic, complement-mediated COVID-19, and could suggest targets for specific thrombotic microvascular injury syndrome with sus- intervention. tained activation of the AP and LP cascades. Poten- tial mechanisms of complement activation, including involvement of positive feedback loops with the INTRODUCTION coagulation system, are discussed in the context of The severe acute respiratory distress syndrome-asso- hypothesis-generating studies providing the founda- ciated coronavirus-2 (SARS-CoV-2), etiologic agent of tion for potential therapeutic intervention. Coronavirus disease 2019 (COVID-19), was initially identified in Wuhan, Hubei, China in December 2019.1 It was documented to be pandemic by the World Health Organization in early March 2020,2 and by METHODS early April there were over 1.5 million cases world- Patient population. All 5 patients were selected for wide, with over 90,000 deaths.3 Organ dysfunction, pathologic studies based on a respiratory tract sample particularly progressive respiratory failure and a gener- positive for SARS-CoV-2 in a reverse transcriptase- alized coagulopathy, are associated with the highest polymerase chain reaction assay, as tested by a desig- mortality.1,4,5 nated diagnostic laboratory. They represent the first 2 It was soon recognized that SARS-CoV-2 is but one patients succumbing to COVID-19 and undergoing of a large pool of prepandemic SARS-like bat corona- autopsy that were available to us, and the first 3 viruses which replicate in primary human airway epi- infected individuals for whom a dermatologic consult thelial cells.6 These include the etiologic agents of the had been requested to evaluate an extensive skin original SARS-CoV and Middle East respiratory syn- rash. These 5 cases were assembled over a period drome (MERS)-CoV, for which mortality is also linked of 2 weeks. A respiratory pathogen PCR panel was to severe respiratory failure, with pathologic evidence used to test for other potential pulmonary viral Translational Research Volume 220 Magro et al 3 pathogens, along with standard bacterial and fungal demonstrated bilateral airspace opacities most promi- respiratory cultures, and these tests were unreveal- nent in the peri-hilar distribution. After discussion with ing. Cases 3 and 4 received hydroxychloroquine his family, he was placed on comfort measures and and/or azithromycin, as noted, with dosing as: died a few hours after presentation. A limited autopsy hydroxychloroquine, 600 mg every 12 hours for 1 was performed. Grossly, the had a congested and day then 400 mg every 12 hours for 4 days; azithro- hemorrhagic appearance. Light microscopic examina- mycin: 500 mg daily for 5 days. tion revealed a severe organizing hemorrhagic pneu- Microscopic and immunohistologic studies. Routine monitis (Fig 1A), including significant fibrin light microscopy and immunohistochemical (IHC) deposition within septal capillary lumens and walls assessment for the deposition of C5b-9 (membrane accompanied by endothelial cell necrosis, consistent attack complex, MAC), C3d, and C4d via a diamino- with a thrombotic necrotizing capillary injury syn- benzidene (DAB) technique was conducted. Identifica- drome (Fig 1B). The pattern of septal capillary injury tion of C5b-9, C3d, or C4d within any epithelial ranged from a pauci-inflammatory pattern (Fig 1C) to basement membrane zone, elastic fibers, or the elastic one characterized by permeation of the interalveolar lamina of vessels was considered nonspecific staining. septa by neutrophils amidst the damaged , IHC was performed using C3d (Cell Marque, Rocklin along with intra-alveolar neutrophils (Fig 1D). No viral CA, 403A-78), C4d (Alpco, Salem NH, BI-RC4d), cytopathic changes were observed. Hyaline membranes C5b-9 (Agilent, Santa Clara CA, M077701-5), and reflective of DAD were not observed. In addition, and MASP2 (Sigma, St. Louis, MO, HPA029313) antibod- in contrast to preliminary reports of COVID-19 lung ies on paraffin embedded sections using a modified pathology,8,9 type II pneumocyte hyperplasia was not Leica protocol. Heat mediated antigen retrieval with appreciated (Fig 1AÀD). The brunt of the lung injury Tris-EDTA buffer (pH = 9, epitope retrieval solution 2) was restricted to septal capillaries, without pneumocyte was performed for 20 minutes, followed by incubation involvement. with each antibody for 15 minutes. Details have been Extensive C4d deposition localized to the interalveo- described by one of the authors, including documenta- lar septal capillaries was then demonstrated (Fig 2A tion of the involvement of many of these complement and B). C5b-9 deposition was also seen, showing a components in pathologic processes linked to comple- similar pattern of septal capillary localization as for ment activation in both dermal and lung tissues.12À15 C4d, although reduced in intensity (Fig 2C). Granular Staining for SARS-CoV-2 spike (catalogue #3525) and deposition of C3d was also noted; although minimal in envelope (catalogue #3531) proteins (Prosci, Poway, comparison to the C5b-9 and C4d deposits, it expressed CA) was optimized with cells from nasopharyngeal a similar pattern of septal capillary localization (Fig preparations obtained from individuals known to be 2D). MASP2 staining was attempted but, due to high positive or negative for the virus before being background staining in this case, was deemed techni- employed in the current studies. It involves use of cally unsatisfactory. horseradish peroxidase conjugate (ENZO, catalogue In order to explore possible generalized complement #ADI-950-113-0100) and an HRP-conjugated compact activation in this patient, a sample of clinically-appear- polymer system, with DAB as chromogen. Hematoxy- ing normal skin was also found to have significant vas- lin and eosin counterstain was used and mounted with cular deposits of C5b-9 within dermal capillaries (not Leica Micromount. shown). Case 2. A 73-year-old male with a history of smok- ing, obesity, and prediabetes developed respiratory dis- tress and was evaluated in the ED. He was febrile, RESULTS tachypneic, severely hypoxemic, and required emer- Case 1. A 62-year-old male with a history of coro- gent endotracheal intubation. He had a serum creati- nary disease, diabetes mellitus, failure with nine of 2.4. Chest x-ray showed bilateral airspace preserved ejection fraction, prior treatment for hepatitis opacities. Sedatives and neuromuscular blockade were C virus infection, and end-stage renal disease on inter- used to enforce a lung protective low-stretch strategy. mittent hemodialysis was evaluated in an emergency His course was complicated by atrial fibrillation, shock, department (ED), in extremis and obtunded. He had and progressive renal failure. His INR and PTT were severe hypoxemia and a of 180/100. within normal limits. On the 4th day of mechanical His international normalized ratio (INR) was mildly ventilation the patient developed thrombocytopenia elevated at 1.2, with a partial thromboplastic time (platelets 148 £ 109/L) and severe hypercapnia, with a (PTT) within the normal range (23.0À37.0), and nor- dramatic increase in the estimated pulmonary dead mal platelet count (178 £ 109/L). Chest x-ray space fraction, and expired the following day. A Translational Research 4 Magro et al June 2020

Fig 1. Microscopic features of pulmonary autopsy samples from Case 1. A, Significant fibrin deposition within the interalveolar septa and alveolar spaces, accompanied by marked hemorrhage and hemosiderin deposition. (Hematoxylin and eosin stain, 200£). B, Prominent destructive septal capillary injury is apparent, with fibrinoid necrosis of the capillaries accompanied by evidence of vascular compromise, with hemorrhage and fibrin and hemosiderin deposition within alveolar spaces. (Hematoxylin and eosin stain, 400£). C, Septal capillary injury is a pauci-inflammatory response. (Hematoxylin and eosin, 1000£). D, The septal capillary injury includes an interstitial and intra-alveolar accumulation of neutrophils. (Hematoxylin and eosin, 400£).

Fig 2. Immunohistochemistry analysis of pulmonary autopsy samples from Case 1. A, Extensive C4d deposi- tion is seen throughout the lung parenchyma, with striking septal capillary localization. (Diaminobenzidene stain, 200£). B, Higher power magnification documents a clear localization of C4d within septal capillaries. (Diaminobenzidene, 1000£). C, A similar septal capillary distribution for C5b-9 deposition is observed, although it is less pronounced than that observed for C4d. (Diaminobenzidene, 1000£). D, A similar septal cap- illary distribution of C3d staining is observed, although it is also less pronounced than was observed for C4d. (Diaminobenzidene, 1000£). Translational Research Volume 220 Magro et al 5 limited autopsy demonstrated a pattern of lung injury tracheal soft tissues, while C4d was localized to that mirrored Case 1. Grossly, the lungs had a hemor- interalveolar septa in regions of microvascular injury rhagic and congested appearance. Microscopically, the (Fig 4C and D). MASP2 staining demonstrated gran- lungs showed extensive hemorrhagic pneumonitis. The ular and punctate staining localized to the interalveo- interalveolar septa were congested, with luminal and lar septa (Fig 4E). mural fibrin deposition within septal capillaries. Focal Case 3. A 32-year-old male with a medical history of intra-alveolar collections of neutrophils and monocytes obesity-associated sleep apnea and anabolic steroid were seen. Viral cytopathic changes were not appreci- use, currently taking testosterone, presented with a 1 ated. Due to vascular compromise, there was concomi- week history of fever and cough. He became progres- tant striking red cell extravasation found within the sively more dyspenic with fevers to 40˚C, ultimately alveolar spaces, along with intra-alveolar fibrin deposi- becoming ventilator dependent from acute respiratory tion. Unlike Case 1, there was some focal hyaline failure. Chest x-ray showed bilateral airspace opacities. membrane formation and type II pneumocyte hyperpla- He had an elevated d-dimer of 1024 ng/ml (normal sia in areas of hemorrhagic pneumonitis (Fig 3AÀD). range 0À229) on presentation, which peaked at However, unlike Case 1, this patient had been on venti- 2090 ng/ml on hospital day 19, and a persistently ele- lator support for 1 week, which can itself lead to some vated INR of 1.6À1.9, but a normal PTT and platelet DAD, and yet, as in Case 1, the dominant pattern was count. Serum complement levels for CH50 (177 CAE septal capillary injury, not DAD. Units, normal range 60À144), C4 (42.6 mg/dL, normal On IHC there was prominent deposition of C5b-9 range 12À36), and C3 (178 mg/dL, normal range within the microvasculature of the interalveolar 90À180) were elevated. Over his continuing 3-plus septa as well as in larger caliber vessels of the lung weeks on ventilator support he completed courses of parenchyma (Fig 4A and B). Of interest, C5b-9 sep- hydroxychloroquine and azithromycin, followed by the tal deposition was not limited to areas of diseased experimental anti-CoV agent remdesivir (5 mg/kg i.v. lung, but was also seen in normal-appearing lung and daily for 10 days).

Fig 3. Microscopic features of pulmonary autopsy samples from Case 2. A, Extensive hemorrhagic pneumonitis was seen, with red cell extravasation and fibrin in alveolar spaces and luminal and mural fibrin deposition within septal capillaries. (Hematoxylin and eosin, 200£). B, The septa exhibit a pauci-cellular pattern of capillary injury as evidenced by significant fibrin deposition, with thrombi seen in capillaries. There is red cell extravasa- tion in the alveolar spaces along with collections of neutrophils and monocytes. (Hematoxylin and eosin, 400£). C, There is slight widening of the septa by a few inflammatory cells, predominantly neutrophils. There is evidence of capillary injury characterized by fibrin deposition in the lumens and walls with red cell extravasa- tion within the septa and adjacent alveolar space. Type II pneumocyte hyperplasia and viral cytopathic effects are not discernible. (Hematoxylin and eosin, 400£). D, Higher power examination further illuminates capillary wall disruption accompanied by fibrin deposition and red cell extravasation, with neutrophils in the septa and within the alveolar spaces. (Hematoxylin and eosin, 1000£). Translational Research 6 Magro et al June 2020

Fig 4. Immunohistochemistry analysis of pulmonary autopsy samples from Case 2. A, There was striking depo- sition of C5b-9 within the microvasculature of the interalveolar septa. (Diaminobenzidene, 200£). B, Higher power magnification again shows localization of C5b-9 within the septa, including C5b-9 deposits in areas of normal appearing lung, suggestive of systemic complement activation. (Diaminobenizdene, 1000£). C, C4d deposition was largely localized to the interalveolar septa in areas of microvascular injury. (Diaminobenzidene 400£). D, A higher power image demonstrates the extensive degree of C4d deposition within the septa. (Diami- nobenzidene, 1000£). E, MASP2 staining showed granular and punctate deposits localized to the interalveolar septa.

After only 4 days on ventilator support, retiform pur- extensive necrosis of the epidermis and adnexal struc- pura with extensive surrounding inflammation was tures, including the eccrine coil. There was a signifi- noted on his buttocks (Fig 5A). Skin showed a cant degree of interstitial and perivascular neutrophilia striking thrombogenic vasculopathy accompanied by with prominent leukocytoclasia (Fig 5B). IHC showed

Fig 5. Clinical, microscopic, and immunhohistochemical analyses of Case 3. A, Striking retiform purpura with surrounding inflammation was noted on the buttocks. B, Skin biopsy showed an extensive pattern of pauci- inflammatory vascular thrombosis with endothelial cell injury. (Hematoxylin and eosin, 400£). C, Prominent deposits of C5b-9 are seen within the microvasculature. (Diaminobenzidene, 400£). Translational Research Volume 220 Magro et al 7 striking and extensive deposition of C5b-9 within the with complete infarction of the area supplied by the left microvasculature (Fig 5C). middle cerebral artery. Case 4. A 66-year-old female, with no significant past Case 5. A 40-year-old female, with no significant medical history, was brought to the ED after 9 days of past medical history, presented to the ED after 2 weeks fever, cough, , and chest pain. She was hypox- of dry cough, fever, myalgias, diarrhea, and progres- emic, with diffuse bilateral patchy airspace opacities, sive dyspnea. She had been diagnosed 1 week earlier, without effusions, on chest x-ray. She was admitted and at an outside hospital, with COVID-19. Echocardio- treated with hydroxychloroquine and prophylactic anti- gram revealed severely reduced left ventricular coagulation with enoxaparin. Three days later she function. Patchy bilateral airspace opacities were noted became confused, increasingly hypoxemic with rising and she was soon intubated for respiratory failure and serum creatinines, and was intubated. Renal replacement shock. D-dimer was elevated at 1187 ng/ml, with a nor- was initiated. On hospital day 10, thrombocytopenia mal platelet count and PTT, but an elevated INR of 1.4. (platelets 128 £ 109/L) and a markedly elevated d-dimer Mildly purpuric reticulated eruptions on her chest, legs of 7030 ng/ml, but normal INR and PTT, were noted. and arms, consistent with livedo racemosa, were noted The next day dusky purpuric patches appeared on her (Fig 7A), and a skin biopsy performed. There was a palms and soles bilaterally (Fig 6A). A skin biopsy of modest perivascular lymphocytic infiltrate in the super- one showed superficial vascular ectasia and an ficial dermis along with deeper seated small thrombi occlusive arterial within the deeper reticular within rare venules of the deep dermis, in the absence dermis in the absence of inflammation (Fig 6B). Exten- of a clear (Fig 7B). Significant vascular sive vascular deposits of C5b-9 (Fig 6C), C3d, and C4d deposits of C5b-9 (Fig 7C) and C4d (Fig 7D) were (Fig 6D) were observed throughout the dermis, with observed. As for Case 4, a biopsy of normal deltoid marked deposition in an occluded artery. A biopsy of skin showed microvascular deposits of C5b-9 through- normal-appearing deltoid skin also showed conspicuous out the dermis (Fig 7E). microvascular deposits of C5b-9 (Fig 6E). Sedative infu- Co-localization of SARS-CoV-2 envelope proteins with sions were discontinued that day, unmasking a comatose complement components in dermal and pulmonary state. Computerized tomographic imaging of the head microvessels of 2 COVID-19 patients. SARS-CoV-2 spike revealed multifocal supra- and infratentorial infarctions, and envelope proteins strongly localized to the

Fig 6. Clinical, microscopic, and immunhohistochemical analyses of Case 4. A, Prominent livedo on the palmar and plantar aspects of the hands and feet, respectively, were noted. B, Skin biopsy demonstrated an occlusive arterial thrombus within deeper dermis. (Hematoxylin and eosin, 200£). C, Extensive endothelial and subendothelial deposits of C5b-9 are observed within the thrombosed artery. (Diaminobenzidene, 400£). D, A similar striking pattern of endothelial and subendothelial C4d deposition is noted within the artery. (Diamino- benzidene, 400£). E, A biopsy of normal-appearing deltoid skin showing conspicuous microvascular deposits of C5b-9. (Diaminobenzidene, 400£). Translational Research 8 Magro et al June 2020

Fig 7. Clinical, microscopic, and immunhohistochemical analyses of Case 5. A, A lacey livedoid rash on the lower extremities was noted. B, A skin biopsy revealed a few deep-seated venules at the dermal-subcuticular interface containing small fibrin thrombi. (Hematoxylin and eosin, 400£). C, There are significant vascular deposits of C5b-9 within the dermis. (Diaminobenzidene, 400£). D, Vascular deposits of C4d were also observed within the dermal microvasculature. (Diaminobenzidene, 400£). E, A biopsy of normal-appearing del- toid skin shows microvascular deposits of C5b-9. (Diaminobenzidene, 400£). respiratory epithelia and the interalveolar septa in tis- in the skin of 3 cases with retiform and purpuric sue from Case 1. No signal was found in control lung lesions, with C5b-9 and C4d deposition in samples samples (data not shown). Using NUANCE software, taken from both cutaneous lesions and normal-appear- by which the fluorescent signal tagging CoV proteins ing skin. appears red and the DAB chromogen tagging C4d Our histologic findings are consistent with emerging appears green, a merged image shows a strong yellow observations suggesting that COVID-19 has clinical signal indicative of septal capillary C4d and SARS- features distinct from typical ARDS. That is, COVID- CoV-2 co-localization (Fig 8AÀD). A similar analysis 19-related severe respiratory distress can be manifest in which the DAB chromogen was used to tag C5b-9 by relatively well-preserved lung mechanics, despite also gave a merged image with a strong yellow signal, the severity of hypoxemia, characterized by high respi- indicative of septal vascular C5b-9 co-deposition with ratory compliance, high shunt fraction, and prolonged SARS-CoV-2 protein (Fig 8EÀH). In terms of the requirement for mechanical ventilation.10,11 The skin, co-localization of C4d with SARS-CoV-2 glyco- pathology in these cases might therefore be expected to protein was we demonstrate for Case 3 (Fig 9A and B). differ from the diffuse alveolar damage and hyaline membrane formation which are hallmarks of typical ARDS. Albeit preliminary pathology studies of lungs from COVID-19 cases described DAD with edema, DISCUSSION hyaline membranes, and inflammation, followed by Using pulmonary and cutaneous biopsy and autopsy type II pneumocyte hyperplasia, features characteristic samples from 5 individuals with severe COVID-19, we of typical ARDS,8,9 the pulmonary abnormalities in document that at least some SARS-CoV-2-infected our patients appear largely restricted to the alveolar patients who become critically ill suffer a generalized capillaries, that is, more of a thrombotic microvascular thrombotic microvascular injury. Such pathology injury with few signs of viral cytopathic or fibroproli- involves at least the lung and skin, and appears medi- ferative changes. An increase in the dead space fraction ated by intense complement activation. Specifically, might be anticipated with this type of pathology, i.e., we found striking septal capillary injury accompanied respiratory failure accompanied by greater lung com- by extensive deposits of the terminal complement com- pliance and less pulmonary consolidation than is char- plex C5b-9 as well as C4d and MASP2 in the lungs of 2 acteristic of typical ARDS. Indeed, the histologic cases examined, and a similar pattern of pauci-inflam- pattern of pauci-cellular terminal lung parenchymal matory complement mediated microthrombotic disease injury with septal capillary damage we document Translational Research Volume 220 Magro et al 9

Fig 8. Demonstration of co-localization of complement components with SARS-CoV2 spike glycoprotein in the lung of Case 1. A, Striking deposition of C4d within the interalveolar septa of the lung was first demonstrated by DAB staining. Using NUANCE software the C4d image appears green (B) while the SARS-CoV2 spike protein appears red (C). D, A merged image shows a significant degree of C4d and SARS-CoV2 co-localization, as revealed by intense yellow staining. EÀH, A similar pattern was observed using an anti-C5b-9 reagent whose image appears green, with a significant degree of C5b-9 and SARS-CoV2 co-localization, as revealed by intense yellow staining. resembles images captured in early case reports in the deposition, might be anticipated,17 as observed in our Chinese language literature of severe COVID-19 cases. It is also consistent with the very high d-dimer pneumonitis.16 levels found in the 3 cases in which it was assessed. We now show that this pathologic pattern, atypical Vascular deposition of C5b-9 is a key feature of many for classic ARDS, is accompanied by extensive deposi- microthrombotic syndromes, regardless of the particu- tion of complement components within the lung septal lar syndromic complex, including catastrophic anti- microvasculature. With such extensive complement phospholipid antibody syndrome, atypical hemolytic involvement, membrane attack complex-mediated uremic syndrome, purpura fulminans and severe multi- microvascular endothelial cell injury and subsequent organ malignant atrophic papulosis, and they may activation of the clotting pathway, leading to fibrin respond to anticomplement therapies.18À23 Translational Research 10 Magro et al June 2020

in limited series of COVID-19 patients from China26 or the United States,27 although there are insufficient numbers to assess its correlation with severity of dis- ease. We have observed an increase in peripheral blood neutrophil vacuolization and granule content in our patients, consistent with an activated state, even if absolute numbers are not elevated. Systemic activation of complement was reflected by elevated complement levels in the sera of SARS-CoV- infected wild-type mice. We measured complement levels in only one of our cases, Case 3. They were only moderately elevated, perhaps reflecting consumption in tissues, with reciprocal depression of circulating levels. Indeed, we found deposition of AP and LP components in normal-appearing skin in 3 cases, not only in biop- sies of retiform cutaneous lesions from those patients. Retiform purpura reflects the cutaneous manifestation of an occlusive microthrombotic process.28 In sum- mary, while complement does not appear to play a major role in controlling CoV replication, it may have a critical role in its pathogenicity. There are several intriguing questions that remain to be addressed as additional cases are investigated. For example, one might have anticipated thrombocytopenia Fig 9. Demonstration of co-localization of C4d and SARS-CoV2 in the context of a systemic microvascular thrombosis spike glycoprotein in the skin of Case 3. A, The skin biopsy was in our cases. Although mild thrombocytopenia stained for C4d showing significant vascular localization (DAB (100À150 £ 109/L) has been recorded in 20% of stain). B, Using NUANCE software C4d is highlighted green while 26 COVID-19 spike protein shows a red staining pattern; a yellow signal COVID-19 patients, such values were seen in only 2 is discernible indicative of co-localization of C4d and viral protein of our 5 patients, nor were platelets a prominent com- within the microvasculature. ponent of the fibrin microthrombi seen histologically. It is possible that infection-related thrombocytosis was A preclinical model of SARS infection emphasizes mitigated by the development of severe vasculopathy. the potential role of complement in the parenchymal However, there is also precedent for such platelet val- lung injury of coronavirus infection.7 Mouse-adapted ues in an atypical hemolytic-uremic syndrome À À SARS-CoV MA15 infection of C3 / mice led to sig- (aHUS)-type of . Platelet counts may nificantly less weight loss and respiratory dysfunction be in the normal range despite severe thrombotic dis- than seen in wild-type mice, and this occurred despite ease.29 Second, there appears to have been a delay of equivalent viral loads in the lung.7 Another feature 5À9 days in most, if not all, of our patients between reflective of a prominence of complement-mediated the onset of typical respiratory symptoms, including a pathology rather than simply inflammation or vasculitis nonproductive cough, myalgias, fatigue, fever, and in our COVID-19 patients was the extent of septal and mild dyspnea, and a catastrophic change in respiratory intra-alveolar neutrophilia. It was present, but pauci- status. This may reflect an evolution of immune pro- inflammatory. SARS-CoV-infected wild-type mice cesses—antiviral IgM levels would expect to emerge at also had higher levels of neutrophils in the lung than this point—or perhaps the fact that the complement À À their C3 / counterparts.7 Tissue neutrophilia may be cascade is a threshold pathway. When activated to a attributable to the neutrophil chemoattractant proper- great extent it may exceed the capacity of complement ties of complement. Both neutrophils and complement regulatory proteins, both soluble and normally present are key sentinels of innate immunity and also modulate in abundance on the microvasculature.29 That raises thrombogenic pathways, the latter thought related to the issue of why only a subset of SARS-CoV2-infected C5a receptor/tissue factor cross-talk mediated by neu- patients develops such severe disease with features trophils.24 Neutrophilia in human SARS-CoV patients atypical for ARDS. The age range was wide among our is associated with a poor outcome and could be an cases, as was the degree of pre-existing immune sup- index of the extent of complement activation.7,25 pression. But there are a variety of complement regula- Peripheral neutrophilia is not a significant observation tory factor polymorphisms or mutations21 as well as Translational Research Volume 220 Magro et al 11 coagulation pathway mutations which could promote to activation of the AP and LP with SARS-CoV-2 spike susceptibility to enhanced complement activation and glycoproteins are consistent with this hypothesis. thrombosis in a given individual that should be investi- There may also be pathways apart from virus spike gated. engagement by which LP and AP are activated. First, Our results, particularly when viewed in the context MBL and ficolins bind to altered glycan structures of murine models of SARS-CoV infection, suggest the present on injured cells, enabling complex formation hypothesis that intervention with inhibitors of the alter- with MASP2.36 In addition, SARS-Cov1 and SARS- native or lectin complement pathways, or both, may be CoV use Angiotensin Converting Enzyme (ACE2) as relevant to severe COVID-19 in humans. Multiple an entry point to cells.6 Angiotensin I and angiotensin studies have documented a role for the AP in microvas- II have been associated with inflammation, oxidative cular injury with C5b-9 deposition,13,30,31 but the lectin stress, and fibrosis, and ACE2 is involved in their deac- pathway has not been similarly investigated. Support tivation.37 If overwhelming coronavirus infection, with for the involvement of the LP in COVID-19 comes binding to ACE2 on epithelial targets not only in the from the discovery that MBL binds to the SARS-CoV lung but in other tissues expressing these proteins, spike glycoprotein.32 A complex of MBL with MASP2 including the , intestines, and brain, were to is the first step in LP activation, and part of a positive interfere with ACE2 activity, the resulting increases in feedback loop leading to sustained AP activation, angiotensin II could lead to reactive oxygen species with inflammation and concurrent activation of the formation and interference with antioxidant and vaso- coagulation cascade.33,34 Although documented experi- dilatory signals such as NOX2 and eNOS, with further mentally for SARS-CoV, this binding is speculative complement activation. This has been demonstrated in in terms of SARS-CoV-2, as certain sites on its glyco- a rat model of angiotensin II overexpression linked to protein spikes are mutated compared to SARS-CoV.35 other disorders.38 The potential loss of auto-vasocon- However, glycosylation sites for high-mannose structures striction and regulation of lung blood flow through with the potential to similarly engage MBL, and thus acti- injured vascular segments would also lead to increased vate MASP2, have been identified for SARS-CoV-2.35 shunting and severe hypoxemia. Fig 10 summarizes Our data showing co-localization of products linked the potential mechanisms of SARS-CoV2 associated

Fig 10. Model for AP and LP complement activation by SARS-CoV2, and its interaction with coagulation cascades. Translational Research 12 Magro et al June 2020 complement activation, and its possible interaction in Wuhan, China. JAMA 2020. https://doi.org/10.1001/ with coagulation pathways. jama.2020.1585. Given these data, use of agents that block LP activa- 5. Tang N, Li D, Wang X, Sun Z. Abnormal coagulation parame- ters are associated with poor prognosis in patients with novel tion such as narsoplimab (OMS721), a human mono- coronavirus pneumonia. J Thromb Haemost 2020. https://doi. clonal antibody against MASP2 which recently org/10.1111/jth.14768. received Breakthrough Therapy Designation for 6. Phan T. Novel coronavirus: from discovery to clinical diagnos- hematopoietic stem cell transplant-linked thrombotic tics. Infect Genet Evol 2020;79:104211. https://doi.org/10.1016/ microangiopathies,39 or , a humanized anti- j.meegid.2020.104211. 7. Gralinski LE, Sheahan TP, Morrison TE, et al. Complement acti- C5 monoclonal antibody FDA-approved for aHUS, vation contributes to severe acute respiratory syndrome corona- might be considered on a case-by-case basis in severe virus pathogenesis. mBio. 2018;9:e01753–18. COVID-19. Investigating critical biomarkers consis- 8. Xu Z, Shi L, Wang Y, et al. Pathologic findings of COVID-19 tent with complement-mediated microvascular injury associated with acute respiratory distress syndrome. Lancet and thrombosis in COVID-19 patients would be impor- Respiratory Med 2020;8:420–2. 9. Zhang H, Zhou P, Wei Y, et al. Histopathologic changes and tant. A panel to study might include: d-dimers; factor SARS-Cov-2 immunostaining in the lung of a patient with VIII, fibrinogen, and other coagulation factors; anti- COVID-19. Ann Intern Med 2020. https://doi.org/10.7326/M20- phospholipid antibodies; C-reactive protein; pro- 0533. inflammatory cytokines, particularly IL-1 and IL-6; cir- 10. Zhou F, Yu T, Du R, et al. Clinical course and risk factors culating complement proteins including C3, C4, C5b-9, for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet 2020;395: and Bb (the latter remaining in circulation longer than 1054–62. other components, as a marker of AP activation); and 11. Gattinoni L, Coppola S, Cressoni M, Busana M, Chiumello D. tissue biopsy, the skin being most readily accessible. Covid-19 does not lead to a “Typical” acute respiratory distress This could enable establishment of criteria for clinical syndrome. Am J Respir Crit Care Med 2020. https://doi.org/ trials with anti-complement and/or anticoagulants, and 10.1164/rccm.202003-0817LE. 12. Magro CM, Poe JC, Kim C, et al. Degos disease: a C5b-9/inter- perhaps open the possibility for earlier intervention feron-a-mediated endotheliopathy syndrome. Am J Clin Pathol than at the end-stages of severe COVID-19. 2011;135:599–610. 13. Magro CM, Momtahen S, Mulvey JJ, Yassin AH, Kaplan RB, Laurence JC. The role of the skin biopsy in the diagnosis of atyp- ical hemolytic uremic syndrome. Am J Dermatopathol ACKNOWLEDGMENTS 2015;37:349–59. The authors are aware of the journal’s authorship 14. Magro CM, Pope Harman A, Klinger D, et al. Use of C4d as a statement and only Dr. Laurence has a potential con- diagnostic adjunct in lung allograft . Am J Transplant flict of interest. He has received an unrestricted edu- 2003;3:1143–54. cational grant and consulting fees from Omeros, Inc. 15. Magro CM, Deng A, Pope-Harman A, et al. Humorally mediated posttransplantation septal capillary injury syndrome as a com- He has recent grants and consulting fees in the recent mon form of pulmonary allograft rejection: a hypothesis. Trans- past from Alexion, Inc. He is also Senior Scientist plantation 2002;74:1273–80. for Programs at amfAR, The Foundation for AIDS 16. Yao XH, Li TY, He ZC, et al. A pathological report of three Research, and has funding from the Angelo Donghia COVID-19 cases by minimally invasive autopsies. Zhonghua Foundation. bing li xue za zhi = Chinese J Pathol 2020. https://doi.org/ 10.3760/cma.j.cn112151-20200312-00193. The expert technical assistance of Bing He, Weill 17. Chaturvedi S, Braunstein EM, Yuan X, et al. Complement activ- Cornell Medicine, Department of Pathology and Labo- ity and complement regulatory gene mutations are associated ratory Medicine, Translational Research Program, is with thrombosis in APS and CAPS. Blood 2020. https://doi.org/ gratefully acknowledged. 10.1182/blood.2019003863. Conflicts of Interest: All authors have read the jour- 18. Magro CM, Poe JC, Kim C, et al. Degos disease: a C5b-9/inter- feron-a-mediated endotheliopathy syndrome. Am J Clin Pathol nal’s policy on disclosure of potential conflicts of inter- 2011. https://doi.org/10.1309/AJCP66QIMFARLZKI. est and have none to declare. 19. Ruffatti A, Calligaro A, Lacognata CS, et al. Insights into the pathogenesis of catastrophic antiphospholipid syndrome. A case REFERENCES report of relapsing catastrophic antiphospholipid syndrome and review of the literature on ischemic colitis. Clin Rheumatol 1. Zhu N, Zhang D, Wang W, et al. A novel coronavirus from 2019. https://doi.org/10.1007/s10067-019-04888-5. patients with pneumonia in China. 2019. N Engl J Med 2020. 20. Manrique-Caballero CL, Peerapornratana S, Formeck C, Del https://doi.org/10.1056/NEJMoa2001017. Rio-Pertuz G, Gomez Danies H, Kellum JA. Typical and atypi- 2. Cucinotta D, Vanelli M. WHO declares COVID-19 a Pandemic. cal hemolytic uremic syndrome in the critically Ill. Crit Care Acta Biomed 2020. https://doi.org/10.23750/abm.v91i1.93973. Clin 2020. https://doi.org/10.1016/j.ccc.2019.11.004. 3. cdc.gov/coronavirus/2019.ncov 21. Fremeaux-Bacchi V, Fakhouri F, Garnier A, et al. Genetics and 4. Wang D, Hu B, Hu C, et al. Clinical characteristics of 138 hospi- outcome of atypical hemolytic uremic syndrome: a nationwide talized patients with 2019 novel coronavirus-infected pneumonia Translational Research Volume 220 Magro et al 13

French series comparing children and adults. Clin J Am Soc 32. Zhou Y, Lu K, Pfefferle S, et al. A single asparagine-linked gly- Nephrol 2013;8:554–62. cosylation site of the severe acute respiratory syndrome corona- 22. Magro CM, Wang X, Garrett-Bakelman F, Laurence J, Shapiro virus spike glycoprotein facilitates inhibition by mannose- LS, Desancho MT. The effects of Eculizumab on the pathology binding lectin through multiple mechanisms. J Virol 2010;84: of malignant atrophic papulosis. Orphanet J Rare Dis 2013. 8753–64. https://doi.org/10.1186/1750-1172-8-185. 33. Beltrame MH, Catarino SJ, Goeldner I, et al. The lectin pathway 23. Perera TB, Murphy-Lavoie HM. Purpura fulminans. StatPearls of complement and rheumatic heart disease. Front Ped Publishing; 2020. Available at: http://www.ncbi.nlm.nih.gov/ 2015;2:1–14. pubmed/30422460 Accessed March 28, 2020. 34. Krarup A, Wallis R, Presanis JS, Gal P, Sim RB. Simultaneous 24. Rits K, Doumas M, Mastellos D, et al. Anovel C5a receptor-tis- activation of complement and coagulation by MBL-associated sue factor cross-talk in neutrophils links innate immunity to serine protease 2. PLoS One 2007;7:e623. coagulation pathways. J Immunol 2006;177:4794–802. 35.WallsAC,ParkY-J,TortoriciMA,WallA,McGuireAT,Veesler 25. Yen Y-T, Liao F, Hsiao C-H, Kao C-L, Chen Y-C, Wu-Hsieh D. Structure, function, and antigenicity of the SARS-CoV-2 spike BA. Modeling the early events of severe acute respiratory syn- glycoprotein. Cell 2020. https://doi.org/10.1016/j.cell.2020.02.058. drome coronavirus infection in vitro. J Virol 2006;80:2684–93. 36. Dodo J, Kocsis A, Gal P. Be on target: strategies of targeting 26. Fan BE, Chong VCL, Chan SSW, et al. Hematologic parameters alternative and lectin pathway components in complement-medi- in patients with COVID-19 infection. Am J Hematol 2020. ated diseases. Front Immunol 2018;9:1–22. https://doi.org/10.1002/ajh.25774. 37. Srivastava P, Badhwar S, Chandran DS, Jaryal AK, Jyotsna 27. Bhatraju PK, Ghassemieh BJ, Nichols M, et al. Covid-19 in criti- VP, Deepak KK. Imbalance between Angiotensin II - Angio- cally ill patients in the Seattle regionÀcase series. N Engl J Med tensin (1-7) system is associated with vascular endothelial 2020. https://doi.org/10.1056/NEJMoa2004500. dysfunction and inflammation in type 2 diabetes with newly 28. Wysong A, Venkatesan P. An approach to the patient with reti- diagnosed . Diabetes Metab Syndr 2019;13: form purpura. Dermatologic Ther 2011;24:151–72. 2061–8. 29. Laurence J, Haller H, Mannucci PM, Nangaku M, Praga M, de Cor- 38. Shagdarsuren E, Wellner M, Braesen JH, et al. Complement acti- doba SR. Atypical hemolytic uremic syndrome (aHUS): essential vation in angiotensin II-induced organ damage. Circ Res aspects of an accurate diagnosis. Clin Adv Hematol Oncol 2016;14 2005;97:716–24. (11S1):1–15. 39. Rambaldi A, Khaled S, Smith M, et al. Improved survival fol- 30. Prufer F, Scheiring J, Sautter S, et al. Terminal complement lowing OMS721 treatment of hematopoietic stem cell trans- complex (C5b-9) in children with recurrent hemolytic uremic plant-associated thrombotic microangiopathy (HCT-TMA). syndrome. Sem Thromb Hemost 2006;32:121–7. Stockholm, Sweden: Eur Hematol Assoc; 2018 June 31. Noris M, Galbusera M, Gastoldi S, et al. Dynamics of comple- 14À17Abst PF724. ment activation in aHUS and how to monitor eculizumab ther- apy. Blood 2014;124:1715–26.